TABLE 3.
Assay Comparison using Reference Antagonists.
| Chemical | Antagonist reference | AR2 antagonist | OT 8 hr | OT 16 hr | UPitt antagonist |
|---|---|---|---|---|---|
| Bicalutamide | Strong | 1 | 1 | 1 | 1 |
| Fenitrothion | Strong | 1 | 1 | 1 | 1 |
| Hydroxyflutamide | Strong | 0 | 1 | 1 | NT |
| Mifepristone | Strong | 1 | 1 | 1 | 1 |
| Spironolactone | Strong | 1 | 1 | 1 | 0 |
| Bisphenol A | Moderate | 1 | 0 | 1 | 1 |
| Cyproterone acetate | Moderate | 0 | 1 | 1 | 1 |
| Flutamide | Moderate | 1 | 1 | 1 | 1 |
| Linuron | Moderate | 1 | 0 | 0 | 1 |
| Prochloraz | Moderate | 1 | 1 | 1 | 1 |
| Fenarimol | Weak | 1 | 0 | 1 | 1 |
| Methoxychlor | Weak | 1 | 0 | 1 | 1 |
| o,p'-DDT | Weak | 1 | 1 | 1 | 1 |
| Procymidone | Weak | 1 | 0 | 0 | 1 |
| Propiconazole | Weak | 1 | 1 | 1 | 1 |
| Vinclozolin | Weak | 1 | 1 | 1 | 1 |
| Zearalenone | Weak | 1 | 0 | 0 | 1 |
| 17-Methyltestosterone | Inactive | 1 | 1 | 1 | 0 |
| 4-Androstene-3,17-dione | Inactive | 1 | 1 | 1 | 0 |
| Atrazine | Inactive | 0 | 0 | 0 | 0 |
| Deltamethrin | Inactive | 0 | 0 | 0 | 0 |
| Methomyl | Inactive | 0 | 0 | 0 | 0 |
| Simazine | Inactive | 0 | 0 | 0 | 0 |
| Testosterone propionate | Inactive | 1 | 1 | 1 | 0 |
| True positive | 15 | 11 | 14 | 15 | |
| False positive | 3 | 3 | 3 | 0 | |
| False negative | 2 | 6 | 3 | 1 | |
| True negative | 4 | 4 | 4 | 7 | |
| Specificity | 57% | 57% | 57% | 100% | |
| Sensitivity | 88% | 65% | 82% | 94% | |
| Balanced accuracy | 73% | 61% | 70% | 97% |
A total of 24 reference antagonists (21 active, 7 inactive) were used to compare the AR2 assay to ToxCast assays from Odyssey Thera (OT) and the University of Pittsburgh (UPitt). The OT, assay was evaluated at two timepoints (8 and 16 h) and has a single protocol that detects both agonists and antagonists. Both the AR2 and UPitt, assays have a distinct antagonist test mode, and both were evaluated at a single timepoint (18 and 2 h, respectively). NT, not tested, 0 = inactive, 1 = active.